Random integers and harmonic analysis (Q948867)

Though many results have been proved on Sidon sets (a set \(\Lambda\) in an abelian discrete group -- say \({\mathbb Z}\) -- is said to be a Sidon set if every continuous function on \({\mathbb T} = {\mathbb R}/{\mathbb Z}\) with spectrum in \(\Lambda\) has an absolutely convergent series; see \textit{D. Li} and \textit{H. Queffélec} [Introduction à l'étude des espaces de Banach. Analyse et probabilités, Cours Spécialisés~12, Société Mathématique de France, Paris (2004; Zbl 1078.46001)]), some important open problems remain open. Two of them are: {\parindent6mm \begin{itemize}\item[1)] the \textit{density conjecture}: does there exist some Sidon set which is dense in the Bohr group (the Bohr compactification of \({\mathbb Z}\))? \item[2)] the \textit{dichotomy conjecture}: does every non-Sidon set be a set of analyticity? \end{itemize}} Recall that a subset \(\Lambda\) of \({\mathbb Z}\) is a set of analyticity if the only functions \(F\) (defined on a neighborhood of \(0\) and null at \(0\)) which operate on \(A (\Lambda) = {\mathcal F} L^1_\Lambda ({\mathbb T})\) (i.e. \(\hat f \in A (\Lambda) \;\Rightarrow \;F(\hat f) \in A (\Lambda)\), whenever \(F(\hat f)\) exists) are the analytic ones. These two problems remain open, though: 1) one has sets dense in the Bohr group which are very close to Sidon sets: they are \(p\)-Sidon for every \(p >1\); see \textit{D. Li}, \textit{H. Queffélec} and \textit{L. Rodríguez-Piazza} [J. Anal. Math. 86, 105--138 (2002; Zbl 1018.43004)]; 2) partial results exist for the dichotomy conjecture: see \textit{J. Bourgain} [A density condition for analyticity of the restriction algebra, Geometrical aspects of functional analysis, Israel Semin. 1985/86, Lect. Notes Math. 1267, 151--156 (1987; Zbl 0625.43006)]. However, these two problems were solved ``statistically'', the first one by \textit{Y. Katznelson} [Suites aléatoires d'entiers, Analyse harmon. Domaine compl., Actes Table ronde internat. Centre Nat. Rech. Sci., Montpellier 1972, Lect. Notes Math. 336, 148--152 (1973; Zbl 0274.42009)] and [Sequences of integers dense in the Bohr group, Proc. Royal Inst. Tech. Stockholm 79--86 (1973)], and the second one by \textit{Y. Katznelson} and \textit{P. Malliavin} [C. R. Acad. Sci., Paris, Sér. A 262, 490--492 (1966; Zbl 0143.36002)]: if we choose integers \(n\) randomly, with expectation \(\delta_n\), then the random set \(\Lambda\) that we get is {\parindent4mm \begin{itemize}\item[--] almost surely a Sidon set which is not dense in the Bohr group, if \(n \delta_n = O (1)\); \item[--] almost surely a set of analyticity (hence not a Sidon set) which is dense in the Bohr group, if \(n \delta_n \to \infty\). \end{itemize}} In the paper under review, the authors revisit these two problems. The main novelty is that they work with independent Poisson random variables \(\xi_n\) and the random set \(\Lambda_\omega = \{ n \in {\mathbb Z}\,;\;\xi_n (\omega) > 0 \}\) instead of independent Bernoulli \((0,1)\)-valued random variables \(\varepsilon_n\), as is usually done. It is the same situation, since \(\Lambda_\omega =\{n \in {\mathbb Z}\,;\;\varepsilon (\omega) = 1 \}\), when \(\delta_n = {\mathbb P}\, (\varepsilon = 1) = {\mathbb P}\, (\xi_n > 0 )\); but this allows the authors to have easier calculations on the martingales introduced for the proof. The paper ends with a discussion on several other open problems on Sidon sets. The results of this paper were announced in \textit{J.-P. Kahane} and \textit{Y. Katznelson} [C. R., Math., Acad. Sci. Paris 345, No.~1, 21--24 (2007; Zbl 1119.43004)]. This paper is followed by another paper: \textit{J.-P. Kahane} and \textit{Y. Katznelson} [Distribution uniforme de certaines suites d'entiers aléatoires dans le groupe de Bohr, J. Anal. Math. 105, 379--382 (2008; Zbl 1153.43005)].